Driver

Computer vision script for simulating gaming racing wheel

Description

When the program is started three windows will open.
All three windows are webcams.
One is at a normal position, one is rotated by 40 degrees and the last one is rotated by -40 degrees.
All three at the same time are trying to detect a person's head each frame of the recording.
If the first normal one detects head then the program knows that person is holding the head at a natural position.
If a camera with a 40-degree angle detects head then the program knows that person tilted head on the right side and presses the ⬅ key on the keyboard.
If a camera with a -40 degree angle detects head then the program knows that person tilted head on the left side and presses the ➡ key on the keyboard.

With haar cascade for head detection I am able to detect persons head on camera.
With three cameras on different angles I am able to detect if head is rotated on left or right or if it is normal.
Depending on which camera detects head (only one out of three cameras can detect face) I am authomaticly pressing ⬅ or ➡ key on keyboard
With that, you can play some racing games

numpy==1.19.2
opencv-python==4.4.0.42
pynput==1.7.1

Webcam, if you dont have webcam you can download DroidCam and use your phone's camera as webcam like I did.

If you want to try this out on desktop games like GTA V you need to uncomment all wk.PressKey() and wk.ReleaseKey() methods comment all keyboard.press() and keyboard.release()
If you choose wk.PressKey() you will press A or D key instead of ⬅➡